Save time and money with secure AI-driven speech to text
TIP Secure Speech provides the most secure, reliable and flexible transcription and translation solutions on the market today.
Delivered for security,
tuned for accuracy.
Our flexible deployment options mean your business can finally integrate powerful AI solutions while maintaining your security requirements.
Trellis Data’s cutting-edge solution is designed to meet the unique needs of public and private sector businesses, providing unparalleled security and control over your data. Our award-winning Speech-to-text delivers industry leading results across a range of languages for both translation and transcription.
Securely translate and transcribe sensitive audio.
We know that correct evidence gathering, proper intelligence, and precise recording of conversations is paramount.
That’s why Trellis Data has developed a proven platform designed to deliver savings in both time and money, all while protecting your sensitive information.
Fast Transcription
AI-powered transcriptions through Secure Speech are produced in real-time.
Accurate transcriptions of important and complex conversations between multiple people can now be capture and reviewed immediately.
Multi-lingual Translation
Languages are automatically recognised and made instantly understandable.
We combine natural language processing and automated speech recognition technology to translate between languages and identify speakers.
Bespoke models
Customisable to improve accuracy in your chosen language or dialect.
Our unique approach to training and tailoring speech-to-text models ensures optimal results for languages that are previously unsupported.
Accelerate your workflow
Adding the power of AI-powered transcription will reduce cost and turnaround times.
Out of the box pre-recorded audio can be transcribed up to 25x faster than real-time for a single Trellis Secure Chat server node. However, Trellis Secure Chat can scale to national levels, and is able to transcribe many thousands of times faster.
| Time | Accuracy | |
|---|---|---|
| Trellis Data TSS | 8 MINS (Per Node) | 95% |
| Manual Transcription | 4 HRS | 99% |
Currently scaled to 3.125 HRS of audio transcription per minute.
Trellis Data commits to accuracy above 95% on average but can achieve 98-100% accuracy in good audio environments, such as court rooms.
Deployable at the edge.
A fully flexible API enabled system
At the edge, inside major assets (such as vehicles), or in secure data centres, TIP Secure Speech can deliver this capability anywhere it is required. Connected to an edge device, such as microphones or cameras as well as other systems, such as intelligence and/or record keeping systems.
The Trellis Data
Universal Translator
Real-time multi-lingual conversations
A fully mobile, end-to-end translation capability enabling two or more speakers to communicate and understand each other across more than 100 languages. Designed to provide near-real time translation for first responders and defence personnel.
Subject speaks into your radio handset or device
Hear the response in your home language
Respond in your home language
Your response will be played in original language
The audio can also be displayed as text and stored for future analysis.
The Trellis Secure Speech difference
-
Turn your conversations into text quickly and seamlessly.
Our language catalogue of highly accurate transcription models includes hundreds of languages that cover more than half of the world’s population. Not only can you transcribe, but you can translate between these same languages in real-time. Rapidly speeding up your processing time.
-
TSS offers numerous ways to capture a variety of new or rare languages and dialects
This includes heavily coded languages like those used by the military or medical sectors. Genuine opportunities exist to customize speech for defense, law enforcement, foreign relations, and healthcare.
-
Optimized for challenging conditions, TSS is the go to AI-tool for the surveillance sector
TSS is proven in both low-volume and noisy environments, with accurate performance in low Signal to Noise Ratio (SNR) scenarios. Producing excellent transcription results with sample rates sub 2kHz, it is also capable of discerning and identifying overlapping voices.
-
Transcription without the need to provide the language.
This is vitally important in multi-lingual scenarios where the spoken language may change several times, and those listening to the audio may not know which language(s) are being used in the conversation.
-
Speaker identification provides several modes designed work for nearly any environment or scenario.
Assign audio to individual speakers within a group where the speakers are not identified (dynamic speaker ID). Recognize specific speakers within a database of known voices. This enables the identification of individuals among numerous discussions. In these cases, TSS will automatically identify the known speaker and label other participants as ‘unknown speaker #’.
-
Map the stress levels of conversations across your organization.
With our emotional analysis models, it’s now possible to quickly understand whether your person of interest is overwhelmed, angry, anxious, or excited. This insight is also particularly valuable for sectors such as law enforcement, where this additional data point can help to ascertain whether a suspect is potentially lying about who they are or their intentions.
Why your enterprise needs Trellis Secure Speech
More productivity
Adding the power of AI-powered transcription will reduce cost and turnaround times and enhance the value of your audio data.
Tailor for your use case
We can work with your organization to fine-tune an audio model designed to deliver quality results in your specific environment.
Integrated
Transcriptions and translations can be seamlessly exported to a custom template [PDF, TXT, DOC] and be available an organization’s CMS and email systems. To ensure secure and straight-forward access for users TIP can also be deployed to include Single Sign-On (SSO).
Maintain custody of your data
In order to exceed strict privacy and security requirements TIP can be installed on premise and can operate in a disconnected environment.
Available where you need it
Deployable at the edge, inside major assets, or in secure data centers, TIP Secure Speech can deliver this capability anywhere it is required.
Flexible audio input
TSS provides Streaming via uni-cast (multiple channels in one stream) or multi-cast (multiple channels in a separate stream). It can also ingest multiple types of audio, including stereo or mono-channel audio input, across various file types (.mp3, .m4a, .aac, .wav, + more).
Tailored made for organisations that demand the highest security and privacy requirements.
Like all our applications, Trellis Secure Speech is designed exclusively for companies with the highest standards of data security and privacy. The Trellis Data Universal Translator is an Deployable at the edge, inside major assets, or in secure data centers, TIP Secure Speech can deliver this capability anywhere it is required. As a fully API enabled system, it can be an edge device, such as microphones or cameras as well as other systems, such as intelligence and/or record keeping systems.